Assessment of Seychelles-Chagos Thermocline Ridge in the CMIP6 Coupled Models

Abstract

The Seychelles-Chagos Thermocline Ridge (SCTR), an open ocean upwelling region in the western Indian Ocean, produces a high biological productivity and becomes an important area of fishing ground to surrounding countries. Accurate reproduction of SCTR in the climate model is essential for future climate variability prediction in the Indian Ocean region. In this study, we assess the SCTR simulated by 27 Coupled Model Intercomparison Project Phase Sixth (CMIP6) models and compared to reanalysis data and 25 previous Coupled Model Intercomparison Project Phase Fifth (CMIP5) models. Most of the CMIP6 models tend to produce considerably deeper thermocline depth (represented by the 20 °C isotherm depth, D20) – a common bias in CMIP5 models. This deep bias is associated with the equatorial easterly wind bias, producing noticeably weak Ekman pumping velocity and deepening the thermocline bias in CMIP6 models. Models with this weak Ekman pumping that produce a deeper thermocline depth, tend to have weaker SST interannual variability, possibly due to weaker thermocline feedback. The shallowest thermocline depth in 5oS-10oS band (thermocline dome) is also located in the eastern part of the Indian Ocean in most of the CMIP6 models whereas the dome from observation data is located around 60o E of the southwestern part of the Indian Ocean. Although the annual mean of D20 from the CMIP6 models appears to be similar to the CMIP5 models, the CMIP6 models simulated quite realistic annual mean of sea surface temperature (SST) in the SCTR region. In seasonal timescale, the SST warms from July to March and cools from April to June with overestimation of SST bias in December to March, opposite of the underestimation bias throughout the year in the CMIP5 models. The CMIP6 models that simulate deeper D20 tend to produce a warmer mean state of SST and have eastward dome longitudinal bias. Although the bias is still noticeable, compared to the CMIP5 models, the CMIP6 models are slightly better in simulating the seasonal cycle of SST, the location of thermocline dome, and thermocline depth in the SCTR region, implying the ability of the CMIP6 models to reproduce surface-subsurface interaction.